4.1.11 Methods and functions

Within this document when I refer to a method I mean a method that can also be a function. If I refer to something as function then this function cannot be a method. The nice thing with methods is that they can be daisy changed together and their synatax is cleaner than functions. The method names are reserved words and CANNOT be used as variable names. To see the complete list of methods available on you build use the command ncap2 -f.

     n2=sin(theta) or n2=theta.sin()
     
     n2=sin(theta)^2 +cos(theta)^2 or  n2=theta.sin().pow(2) + theta.cos()^2

The below statement converts three_dmn_var_sht to type double, finds the average, then converts this average back to type short.

     three_avg=three_dmn_var_sht.double().avg().short();

avg()

Mean value

sqravg()

Square of the mean

avgsqr()

Mean of sum of squares

max()

Maximum value

min()

Minimum value

rms()

Root-mean-square (normalized by N)

rmssdn()

Root-mean square (normalized by N-1)

ttl() or total()

Sum of values

     // Average a variable over time
     four_time_avg=four_dmn_rec_var($time);

pack() & pack_short()

The default packing algorithm is applied and variable is packed to NC_SHORT

pack_byte()

Variable is packed to NC_BYTE

pack_short()

Variable is packed to NC_SHORT

pack_int()

Variable is packed to NC_INT

unpack()

The standard unpacking algorithm is applied.

Basic Methods
These methods work with variables and attributes. They have no arguments

size()

Total number of elements

ndims()

Number of dimensions in variable

type()

Returns the netcdf type (see previous section)

set_miss(expr)

Takes one argument the missing value. Sets or overwrites the existing missing value. The argument given is converted if necessary to the variable type

change_miss(expr)

Changes the missing value elements of the variable to the new missing value (n.b. an expensive function).

get_miss()

Returns the missing value of a variable in Input or Output

delete_miss()

Deletes the missing value associated with a variable.

ram_write()

Writes a RAM variable to disk i.e. converts it to a regular disk type variable

ram_delete()

Deletes a RAM variable or an attribute

reverse(dim args)

Reverses the dimension ordering of elements in a variable.

permute(dim args)

Re-shapes variables by re-ordering the dimensions. All the dims of the variable must be specified in the arguments. A limitation of this permute (unlike ncpdq) is that the record dimension cannot be re-assigned.

     lat_2D_rrg_new=lat_2D_rrg.permute($lon,$lat).reverse($lon);
     lat_2D_rrg_new=0,90,-30,30,-30,30,-90,0

netCDF3/4 Types

byte()

convert to NC_BYTE a signed 1 byte integer

char()

convert to NC_CHAR a ISO/ASCII character

short()

convert to NC_SHORT a signed 2 byte integer

int()

convert to NC_INT a signed 4 byte integer

float()

convert to NC_FLOAT a single precision floating point number

double()

convert to NC_DOUBLE a double precision floating point number

netCDF4 Types

ubyte()

convert to NC_UBYTE a unsigned 1 byte int

ushort()

convert to NC_USHORT a unsigned 2-byte int

uint()

convert to NC_UINT a unsigned 4-byte int

int64()

convert to NC_INT64 a signed 8-byte int

uint64()

convert to NC_UINT64 a unsigned 8-byte int *

Intrinsic Mathematical Methods
The list of mathematical methods is system dependant. For the full list see Intrinsic mathematical methods

All the mathematical methods take a single operand ,with the exception of atan2 and pow which take two. If the operand type is less than float then the result will be of type float. If the operand is type double then the result will be type double. Like the other methods, you are free to use the mathematical methods as functions.

     n1=pow(2,3.0f)    // n1 type float
     n2=atan2(2,3.0)   // n2 type double
     n3=1/(three_dmn_var_dbl.cos().pow(2))-tan(three_dmn_var_dbl)^2; // n3 type double